Insights into nano-heterostructured materials for gas sensing: a review

The living standards of humans have grown at a remarkable pace, which was accompanied by environmental issues such as monitoring of air quality [1–3]. It is now crucial to detect and monitor fossil fuel hydrocarbons, oxygen for breathable atmospheres and combustion processes, in addition to other gases that pollute the environment and pose a risk to public health [4–6]. Pollution control, in particular, requires gas sensors to monitor the environment, public security and food quality [7]. With such prominence, there is a strong trend to integrate sensors into devices, especially within the internet of things paradigm [8]. Examples are systems for air quality control inside buildings and heavy traffic regions, and detection of toxic and explosive gases. The gas sensing market is currently worth €630 million annually, with chemiresistive sensors comprising ∼€230 million due to their low cost, high sensitivity, fast response, and relative simplicity [9].

The essential features of a gas sensor are the three 'S' rules: sensitivity, selectivity, and stability. Chemiresistive gas sensors have been optimized with wide-bandgap semiconducting metal oxides, such as tin oxide, tungsten oxide, indium oxide, or zinc oxide [4, 10, 11]. They act as gas sensors when adsorbed gaseous species form surface states in the metal oxide by exchanging electrons with the bulk material. Optimization is mainly focused on decreasing the particle (grain) size of the semiconducting metal oxide to the nanometer scale. In nanoparticles, the surface states extend throughout the entire particle, inducing an increase in sensitivity together with the high surface-to-volume ratio and large accessible surface area. With nanomaterials, microsensor platforms may be developed [12–16] where miniaturization reduces material consumption in manufacture and decreases power consumption during operation [17–19]. Miniaturization also enables faster response times and lower detection limits than traditional ceramic sensors, enabling devices to be deployed on-site and for real-time detection. Chemiresistive sensors made with semiconducting oxides (SnO₂, ZnO, WO₃, In₂O₃), catalytic oxides (V₂O₅, MoO₃, CuO, NiO), and mixed oxides (LaFeO₃, BaTiO₃, and ZnFe₂O₄) have been used in environmental monitoring, personal safety, and industrial processes [20, 21]. They possess drawbacks related to low selectivity, extended response and recovery times, and instability [22], which have been addressed by modifying the host matrix with metal sensitizers, including Ag, Au, Fe, Pd, Pt, Cu, or other metal oxides such as NiO, CuO and SnO₂ [23, 24]. Core–shell structures have also been used, including α-Fe₂O₃/SnO₂, GaN/WO₃, TiO₂/ZnO, multi-walled CNT/SnO₂, ZnO/SnO₂, CeO₂/TiO₂, Fe₂O₃/ZnO, MoO₃/TiO₂ and Fe₂O₃/TiO₂ [25, 26].

The fabrication of heterostructures with p-type and n-type semiconducting oxides permits one to combine physical and chemical properties of individual components. The performance in terms of sensitivity and fast response can then be enhanced owing to the build-up of an inner electric field at the p/n junction. In equilibrium, the inner electric field in the depletion region sets up a negative charge at the p-type semiconductor, while a positive net charge is established in the n-type semiconducting region. Figure 1 shows an overview of gas sensors made with metal oxide heterostructures (MOH) for detecting toxic analytes such as ethanol, acetone, formaldehyde, ammonia (NH₃), hydrogen sulphide (H₂S), carbon monoxide (CO), nitrogen dioxide (NO₂), carbon dioxide (CO₂), ozone (O₃), methane (CH₄), liquefied petroleum (LPG). Also depicted in figure 1 is a chart showing the increase in publications over the last decade.

Zoom In Zoom Out Reset image size

The design and development of new sensing materials have been the subject of various review papers on sensing mechanisms or target gases. For example, Zappa et al [25] summarized heterostructure-based gas sensors that operated mainly at high temperatures, with emphasis on the physical and chemical syntheses but there was no attempt to identify the approaches more likely to lead to high sensitivity. Miller et al [27] presented a thorough review on gas sensors made with metal oxide heterojunctions where the sensing mechanism was analyzed in depth and the factors affecting the sensing performance were discussed. Similar reviews were produced by Miller et al [27] with focus on the morphology and sensing mechanism of one-dimensional heterostructured materials, and by Walker et al [26] with an emphasis on synergistic effects from oxide heterostructures in gas sensing. All of these reviews focused on particular subjects without covering recent advances in nano heterostructured materials for gas sensors from a pragmatic perspective. This is precisely the focus on the present review, which is aimed at a comprehensive coverage of recent advances on heterostructured materials with the prospects of continued efforts toward the development of higher performance gas sensors.

1.1. General sensing mechanism

Metal oxide gas sensors operate based on a change in the electrical properties of the active element that adsorbs gas molecules, as illustrated in figure 2(a). The negative charge trapped in the adsorbed oxygen species causes an upward band bending and reduced conductivity compared to the flat band situation in n-type materials. When adsorbed on metal oxides, O₂ molecules can extract electrons from the conduction band (E_c), thus trapping electrons at the surface in the form of ions. This will lead to band bending, and an electron depleted region known as a space-charge layer whose thickness is of the order of the band bending region length. The reaction of the oxygenated species with reducing gases, or a competitive adsorption/replacement of the adsorbed oxygen by other molecules decreases and can reverse the band bending, resulting in an increased conductivity [28, 29]. Depending on the temperature, oxygen is mainly adsorbed on the sensor in ionic forms such as O₂ ⁻ (<100 °C), O⁻ (100 °C–300 °C), and O²⁻ (>300 °C). Figure 2(b) shows the changes in band structure owing to the adsorption of oxygen molecules.

Zoom In Zoom Out Reset image size

Metal oxides such as SnO₂, WO_3, and ZnO are n-type materials, with the stoichiometric excess of metal attributed to oxygen vacancies (O₂ ⁻, O⁻ and O²⁻). Under standard sensor operating conditions, the conductivity is lower than that in bulk due to surface oxygen ions that trap electrons. A surface depletion layer is formed, and thus a Schottky barrier is established at interparticle contacts. Such barriers control the injection of charge carriers, and their height is related to the probability of adsorption. Heterojunctions are obtained by modifying the sensor surface with a thin layer of metal oxides, polymers, noble metals, or 2D materials, leading to random heterojunctions distributed uniformly over the sensor surface [32]. The formation of these junctions causes depletion of nanostructures with the formation of potential barriers distributed over the sensor film surface. Table 1 brings details of sensors to detect toxic gases obtained with heterostructures containing ZnO and SnO₂ transition metal oxides.

Table 1. Gas sensing performance of ZnO and SnO₂-based heterostructures for detecting toxic gases.

Analyte gas	Morphology	Composition	Operating temperature	Concentration	Sensor response	Response/recovery time	LOD	Selective against	Reference
Butanol	Mixed particles	ZnO–SnO2	350	5 ppm	300 a	NA h	0.1 ppm	Ethanol, propanol, hexanol, 3-methyl-1-butanol, 3-octanone, diacetyl, butanal, butanone, ethyl benzene, decane	[43]
Ethanol	Nanofibers (NFs)	SnO2–ZnO	360	2500 ppm	17 b	5 s/1 s	27.7 ppm	NA h	[44]
Ethanol	Core–shell nanosheet	SnO2/ZnO	350	100 ppm	13.3 b	NA h	5 ppm	Benzene, toluene, ammonia, acetone, methanol	[45]
Hydrogen sulphide	Hetero-nanostructures	SnO2/ZnO	100	10 ppb	2.7 b	∼150 s/40 s	10 ppb	Ethanol, acetone, NH3, H2	[46]
Ammonia	Composite fiber	ZnO/NiO	RTg	250 ppm	67 b	∼1000 s/—	50 ppm	Methanol, 2-propanol, ethanol, benzyl alcohol, acetone, acetic acid, trimethylamine	[47]
Carbon monoxide	Composite	SnO2–CuO	235	10 ppm	95 b	37 s/80 s	10 ppm	Ethanol, methanol, toluene, acetone, formaldehyde	[48]
Formaldehyde	Hollow spheres	SnO2 /ZnO	RTg	100 ppm	76 c	36 s/73 s	1.91 ppb	Ammonia, ethanol, acetone, methanol, benzene, toluene	[49]
Nitrogen dioxide	Hierarchical nanostructures	SnO2@ZnO	150	0.005 ppm	1.2 d	60 s/45 s	5 ppb	Acetone, C7H8, H2S, CHCl3, NH3, ethanol	[50]
Acetone	Heterostructures	NiO/ZnO	400	100 ppm	6.7 e	10 s/—	11 ppm	NA h	[51]
Triethylamine (TEA)	Composite	SnO2/NiO	70	10 ppm	14.3	86 s/95 s	2 ppm	NO2, HCHO, C2H5OH, CH3OH, C6H5CH3, NH3	[52]
Methanol	Heterostructures	ZnO/SnO2	200	100 ppm	80 b	20 s/65 s	1 ppm	NA h	[53]
Ozone	Heterojunctions	ZnO–SnO2	RT	20 ppb	8 f	13 s/90 s	20 ppb	NO2, NH3, CO	[54]
Chlorine	Heterostructures	ZnO–SnO2	260	10 ppm	230.52 f	1 s/3 s	0.06 ppm	Ammonia, formaldehyde, methanol, acetone, ethanol, NO2	[55]
Acetone	Thick films	ZnO/SnO2	180	0.5 ppm	3.36 b	57 s/63 s	10 ppb	Alcohol, methanol, gasoline, ammonia, CO, CO2	[56]
TEA	Heterostructure	Au–ZnO/SnO2	300	100 ppm	115 b	7 s/30 s	2 ppm	Benzene, p-xylene, acetone, ethanol, isopropanol	[57]
Hydrogen	Composites	ZnO–SnO2	150	1%	65% c	∼200 s/∼ 100s	0.3%	NA h	[58]

^a S = (ΔI/I_a) × 100. ^b S = R_a/R_g. ^c R_s = (R_a − R_g)/R_a × 100. ^d S = (G_f − G₀)/G₀ = ΔG/G₀. ^eResponse = (G_air − G_gas)/G_gas. ^f S = R_g/R_a. ^gRT—room temperature. ^hNA—not available.

Novel materials have been developed to address such limitations of gas sensors as sluggish response and recovery time, low sensitivity and selectivity, and high-temperature operation. In this context, heterostructures are used to improve the performance of pristine nanomaterials, as discussed in review papers [33–35]. They are mostly prepared with wet chemical approaches, including solvothermal, hydrothermal, and co-precipitation methods. In wet chemistry, synthesis occurs with chemical reactions under controlled conditions, above ambient temperature and pressure in an autoclave [36]. Heterostructures demand two or more fabrication steps because different materials are combined, making it difficult to reach mass production, in addition to the possible low-quality heterojunctions [37]. This is the case of heterostructures with 2D materials need mechanical [38], chemical [39], or optical [40] exfoliation steps. An alternative is to employ electrostatic interactions with a magnetic stirrer or hydrothermal procedures [41, 42]. The distinct methods used to synthesize heterostructures for gas sensors are summarized in the following, and some selected examples are given in table 2. The morphologies of heterostructures produced with different synthesis methods are illustrated in figure 3.

Zoom In Zoom Out Reset image size

2.1. Hydrothermal method

2.2. Co-precipitation method

2.3. Solvothermal method

2.4. Electrospinning method (ES)

2.5. Spray pyrolysis (SP)

SP is a simple and practical approach to design nanostructured materials with controlled morphology, suitable for scalable production. Advantages of SP include the fabrication of homogeneous multi-component materials and composites with specific stoichiometric ratios, rapid solvent evaporation, and solvent decomposition porous hollow structures. SP can also be scaled up for mass production. Srinivasan et al [63] synthesized n–n heterostructures of ZnO/CdO (figures 3(g) and (h)) with SP, and used them in room temperature (RT) sensing of acetone gas under UV-light illumination. Some examples of heterostructures synthesized by the SP method are SnO₂/ZnO, SnO₂/Co₃O₄, GO/SnO₂, GO/ZnO, and GO/Co₃O₄ [83].

Because of their controlled and straightforward manner of designing nanostructures, these synthesis approaches are promising for obtaining materials with the desired sensing properties. There is no definitive answer to which method is better to achieve a specific property or detect selective gas because many parameters, such as operating temperature, size, and shape of the morphology, play a significant role [84]. In this regard, researchers have mostly opted for wet chemical approaches, particularly hydrothermal methods, to fabricate heterostructures with porous morphology. Figure 4 indicates the fabrication method appropriate for detecting toxic analytes in a specific temperature range. Also mentioned is the suitability of each method for a target gas. It is worth noting, nevertheless, that the figure brings only a pictorial view based on one parameter. Other sensing characteristics must also be considered, including morphology, size/shape, and surface area, which can affect the sensor response.

Zoom In Zoom Out Reset image size

The formation of heterostructures is a popular way to enhance gas sensing performance because it increases reaction and adsorption sites, yielding a higher catalytic activity than a single material. Combining two different materials such as n–n, p–p, or n–p semiconductors provides synergistic effects of the interfaces, with equalization of the Fermi energies due to electron transfer from higher energies to the unoccupied lower energy level. A potential energy barrier is formed at the interface with band bending due to different Fermi energies. Hence, electrons must cross the depletion region (interface) to overcome this potential energy barrier. The detailed mechanisms behind the operation of gas sensors made of MOHs are discussed in [84, 198–200].

Figure 5 shows the energy band diagram and the heterojunctions formed with n-type and p-type semiconductors. In n–n metal oxide heterojunctions (figure 5(a)), electrons make the majority charge carriers. At the interface, an electron depletion layer is formed with higher-energy conduction band states due to the loss of electrons, in addition to an electron accumulation layer with lower-energy conduction band states [201]. For p–p metal oxide heterojunctions (figure 5(b)), holes are the majority charge carriers that are transported at the interface from higher to lower-energy valence band states. This yields a hole depletion region and a hole accumulation region with higher and lower energy valance band states, respectively. A more widespread use is observed for n-type materials due to their stability in low oxygen environments and response compatibility to measuring devices [202]. Sensing performance can be enhanced with a p–n heterojunction. At a n–p or p–n heterostructure interface (figure 5(c)), electron transfer from n-type to p-type or hole transfer from p-type to n-type occurs due to electron–hole recombination depending on the backbone material. This facilitates additional oxygen adsorption due to the larger electron density [203]. These heterostructures have a larger number of free electrons than in n–n or p–p heterojunctions [27]. The intrinsic Fermi level for n-type materials is usually higher than for a p-type, which results in a space charge region at the p–n interface. The width of this region is calculated using equation (1), where X_n and X_p are electron depletion and hole depletion layers, respectively:

$$\begin{equation}{X_{\text{n}}} = {\left\{ {\frac{{2{\varepsilon _{\text{n}}}{V_{{\text{bi}}}}}}{q}\left[ {\frac{{{N_{\text{p}}}}}{{{N_{\text{n}}}}}} \right]\left[ {\frac{1}{{{N_{\text{n}}} + {N_{\text{p}}}}}} \right]} \right\}^{1/2}}\quad{\text{and}}\quad{\mkern 1mu} {X_{\text{p}}} = {\left\{ {\frac{{2{\varepsilon _{\text{p}}}{V_{{\text{bi}}}}}}{q}\left[ {\frac{{{N_{\text{n}}}}}{{{N_{\text{p}}}}}} \right]{\text{ }}\left[ {\frac{1}{{{N_{\text{n}}} + {N_{\text{p}}}}}} \right]} \right\}^{1/2}}.\end{equation} \tag{ 1 }$$

Zoom In Zoom Out Reset image size

here, _n and _p are the static dielectric constant, N_n and N_p are the carrier concentration of the n-and p-type semiconductor, and qV_bi is the high potential barrier between n and p-type semiconductors. p–n heterostructures provide a high sensor response and selective sensing towards target analytes for several reasons. These include effective charge separation, long lifetime of charge carriers, minimized electron–hole recombination [27, 145, 204–206]. The relevant parameters to evaluate gas sensing performance are the electrical response, gas selectivity and sensitivity, operating temperature, and response and recovery times [27], which will be considered in the brief survey below of gas sensors made with MOHs in recent years.

Heterojunctions can accelerate electron transport and improve oxygen adsorption, resulting in an abundance of oxygen vacancies on the heterostructures surfaces and providing new active sites for sensing reactions. Furthermore, the heterostructures often contain mesoporous that aid in gas adsorption and desorption. These effects should increase the sensitivity and response rates of gas sensors made with MOHs. In the formation of the heterojunction, many defects are generated at the interface between the composites due to the different lattice parameters of the two materials, thus creating active sites for gas adsorption. Therefore, more gas molecules can be adsorbed on the composites and participate in the reaction between the chemically adsorbed oxygen and the target molecule. The gas sensing mechanism can be verified experimentally by measuring the work function using the Kelvin probe method. Figure 6(a) shows the energy diagram for a ZnO@In₂O₃ heterojunction obtained from scanning Kelvin probe force microscopy (SKPM) under NO₂ gas exposure at RT [207]. Considerable changes exist in the energy band structure owing to the local electron transport between ZnO and In₂O₃ at the heterojunctions upon exposure to NO₂ gas, thus indicating charge carrier recombination. Electrons shift from a higher to a lower Fermi energy before the energy levels reach equilibrium. Under environmental air, both n-type materials adsorb negatively charged oxygen ions owing to the oxygen adsorption over the sensing material. As a result, there is a built-in interfacial electric field and electrons can move from the conduction band of In₂O₃ to ZnO. Hence, these adsorbed oxygens ultimately draw electrons and make the n-type In₂O₃ more depleted, with this lower electron concentration yielding an upward band bending. Similarly, n-type ZnO becomes more conductive due to a higher electron concentration causing a downward band bending, as illustrated in figure 6(b). The NO₂ gas molecules trap the electrons by interacting with adsorbed oxygen species on the heterojunction interfaces (figure 6(c)). Since the affinity of NO₂ gas (2.28 eV) is higher than the adsorbed oxygen species (0.43 eV), NO₂ reacts at the heterojunction via oxygen species (O⁻). The ensuing increase in the potential barrier was confirmed in the studies with the Kelvin probe. Figures 6(d) and (e) show the SKPM analysis of the ZnO@In₂O₃ heterostructure before and after exposure to the gas. In air, the contact potential difference (CPD) was 103 mV, while it was 35 mV for a pure ZnO sample. The difference is due to the charge transport between the heterojunction materials, with the electrons moving because of the difference in work functions. The calculated work function for ZnO@In₂O₃ heterostructure was 4.97 eV, higher than for the pure ZnO sample, leading to band bending. In the presence of NO₂ gas, CPD further increased to 146 mV with a work function of 5.02 eV. This increase in CPD depends on the difference in work function, as indicated in the SKPM images of figures 6(d) and (e). The energy band bending could be understood by simulating the energy band structure using the evidence from SKPM with n-ZnO/n-In₂O₃ heterojunctions. Furthermore, the crystallographic defects and increased surface area of ZnO@In₂O₃ nanowires provided favorable adsorption facets for a fast response to gaseous NO₂, potentially amplifying the gas sensing properties.

Zoom In Zoom Out Reset image size

3.1. p–n heterostructures

Various metal oxides have been used in p–n heterostructures for gas sensors, including NiO/ZnO [148, 208, 209], NiO/SnO₂ [210, 211], Co₃O₄/ZnO [82, 212], CuO/ZnO [68, 213–215], SnO/SnO₂ [120], Fe₂O₃/NiO [216] and CuO/TiO₂ [217]. The gas sensing performances of some p–n heterostructures at RT and high temperature are described in table 3. Emphasis has been placed on detecting toxic gases and VOCs, as with ZnO nanosheets and NiO nanoparticles to detect TEA [208, 210]. Using the p–n heterojunction led to a higher sensitivity compared to pristine ZnO. Figure 7(a) shows the energy band diagram and the schematic model of the NiO/ZnO p–n junction in air and acetone gas. A depletion layer is formed on the ZnO surface and an accumulation layer on NiO, exhibiting high electrical resistance (R_a) in dry air. In contrast, upon exposure to acetone gas, the atmosphere with abundant electrons induces the shrink in the depletion layer, causing a change in electrical conductivity (R_g) and, therefore, a high sensing response. Figure 7(b) shows the energy band diagram and the schematic model of the NiO/SnO₂ p–n junction in air and in TEA gas conditions. When the sensor is under air conditions, the resistance of the composite (R_a) is higher than that of the pristine SnO₂, due to the depletion layer in the p–n junction. Then, the composite resistance decreases upon exposure to TEA gas (R_g), improving the sensor response due to the large change in resistance (S = R_a/R_g).

Zoom In Zoom Out Reset image size

Table 3. Gas sensing performance of p–n heterostructures.

Analyte gas	Morphology	Composition	Operating temperature	Concentration	Sensor response	Response/recovery time	LOD	Selective against	Reference
Triethylamine (TEA)	Nanoparticles—nanosheets	NiO/ZnO	320 °C	100 ppm	185 a	7 s/33 s	2 ppm	Ethanol, acetone, 2-propanol, p-xylene, C6H6, C6H12, CH3OH, n-hexane	[208]
Hydrogen sulfide (H2S)	Nanowires	NiO/ZnO	RT k	100 ppm	31.5 b	15 s/10 min	10 ppm	Carbon disulfide, ethanol, acetone, methanol	[209]
Acetone ((CH3)2CO)	Nanoparticles—hollow spheres	NiO/ZnO	275 °C	100 ppm	30 a	1 s/20 s	0.8 ppm	Benzene, formaldehyde, xylene, methanol, ethanol	[148]
Triethylamine (TEA)	Nanoparticles—hollow spheres	NiO/SnO2	220 °C	10 ppm	46.5 a	11 s/34 s	2 ppm	Ethanol, acetone, benzene, xylene	[210]
Formaldehyde (HCHO)	Hollow spheres—nanoparticles	Co3O4/ZnO	160 °C	10 ppm	5.68 c	27 s/15 s	1 ppm	NO2, NH3, ethanol, acetone	[212]
Formaldehyde (HCHO)	Core–shell NFs	Co3O4/ZnO	220 °C	100 ppm	∼5 d	6 s/9 s	10 ppm	Ethanol, acetone, toluene, ammonia, carbon monoxide	[82]
Ethanol (C2H5OH)	Flower-like nanorods	CuO/ZnO	300 °C	100 ppm	98.8 a	7 s/9 s	1 ppm	Formaldehyde (CH2O), methanol (CH3OH), benzene (C6H6), dichloromethane (CH2Cl2), carbon tetrachloride (CCl4)	[213]
Ethanol (C2H5OH)	Core–shell cubes	CuO/ZnO	240 °C	50 ppm	∼3.5 e	5 s/18 s	5 ppm	CH3OH, (CH3)2CHOH, CH3COCH3, CO, CH4	[215]
Ammonia (NH3)	Nanocomposites	Al-CuO/ZnO	RT k	200 ppm	510.11 a	14 s/9 s	50 ppm	Ethanol, acetone, methanol, isopropanol, n-butanol, acetyl acetone, gasoline, xylene	[68]
Nitrogen dioxide (NO2)	Nanosheet crystals	SnO/SnO2	RT k	0.2 ppm	2.5 d	57 s/5 min	0.1 ppm	Ethanol, acetone, benzene, ammonia, toluene	[120]
Ethanol (C2H5OH)	Nanosheets	Fe2O3/NiO	255 °C	100 ppm	170.7 d	0.5 s/14.6 s	200 ppb	Methanol, formaldehyde, DMF (dimethylformamide), DMA (dimethylacetamide), toluene, methanamide, ammonia	[216]
Hydrogen gas (H2)	Thin film—nanotubes	CuO/TiO2	200 °C	1000 ppm	2 f	7.4 min/6.8 min	100 ppm	Acetone, ethanol, chloroform, NO2	[217]
Butanol	Mixed particles	ZnO–SnO2	350 °C	5 ppm	300 g	NA l	0.1 ppm	Ethanol, 1-propanol, 1-butanol, 1-hexanol, 3-methyl-1-butanol, 3- octanone, diacetyl, butanal, 2-butanone, ethyl benzene, decane	[43]
Ethanol	NFs	SnO2–ZnO	360 °C	2500 ppm	17 a	5 s/1 s	27.7 ppm	NA l	[44]
Ethanol	Core–shell nanosheet	SnO2/ZnO	350 °C	100 ppm	13.3 a	NA l	5 ppm	Benzene, toluene, ammonia, acetone, methanol	[45]
Hydrogen sulphide	Hetero-nanostructures	SnO2/ZnO	100 °C	10 ppb	2.7 a	∼150 s/40 s	10 ppb	Ethanol, acetone, NH3, H2	[46]
Ammonia	Composite fiber	ZnO/NiO	RT k	250 ppm	67 a	1000 s/—	50 ppm	Methanol, 2-propanol, ethanol, benzyl alcohol, acetone, acetic acid, trimethylamine	[47]
Carbon monoxide	Composite	SnO2–CuO	235 °C	10 ppm	95 a	37 s/80 s	10 ppm	Ethanol, methanol, toluene, acetone, formaldehyde	[48]
Formaldehyde	Hollow spheres	SnO2 /ZnO	RT k	100 ppm	76 h	36 s/73 s	1.91 ppb	Ammonia, ethanol, acetone, methanol, benzene, toluene	[49]
Nitrogen dioxide	Hierarchical nanostructures	SnO2@ZnO	150 °C	0.005 ppm	1.2 i	60 s/60 s	5 ppb	Acetone, C7H8, H2S, CHCl3, NH3, ethanol	[50]
Acetone	Heterostructures	NiO/ZnO	400 °C	100 ppm	6.7 j	10 s/—	11 ppm	NA l	[51]
TEA	Composite	SnO2/NiO	70 °C	10 ppm	14.3 d	86 s/95 s	2 ppm	NO2, HCHO, C2H5OH, CH3OH, C6H5CH3, NH3	[52]
Methanol	Heterostructures	ZnO/SnO2	200 °C	100 ppm	80 a	20 s/65 s	1 ppm	NA l	[53]
Ozone	Heterojunctions	ZnO–SnO2	RT	20 ppb	8 f	13 s/90 s	20 ppb	NO2, NH3, CO	[54]
Chlorine	Heterostructures	ZnO–SnO2	260 °C	10 ppm	230.52 d	1 s/3 s	0.06 ppm	Ammonia, formaldehyde, methanol, acetone, ethanol, NO2	[55]
Acetone	Thick films	ZnO/SnO2	180 °C	0.5 ppm	3.36 a	57 s/63 s	10 ppb	Alcohol, methanol, gasoline, ammonia, CO, CO2	[56]
TEA	Heterostructure	Au–ZnO/SnO2	300 °C	100 ppm	115 a	7 s/30 s	2 ppm	Benzene, p-xylene, acetone, ethanol, isopropanol	[57]
Hydrogen	Composites	ZnO–SnO2	150 °C	1%	65% h	∼200 s/∼ 100s	0.3%	NA l	[58]
H2S	Microtubes	CuO/ZnO	170 °C	50 ppb	∼1.66	35/29 s	10 ppb	Ethanol, chlorobenzene, acetone, formaldehyde, aniline, TEA, ammonia, hydrogen, nitrogen dioxide, carbon monoxide	[218]
Formaldehyde	Nanoflowers	SnO/SnO2	120 °C	50 ppm	80.9	7/27 s	8.15 ppb	Acetone, ethanol, methanol, TEA	[164]
CO	NFs	SnO2–Cu2O	300 °C	1 ppm	3	17/12 s	NA l	C6H6, NH3, C3H3O, and HCHO	[111]
H2S	Hierarchical	NiO@SnO2	240 °C	50	183.5	45/60 s	1.5 ppb	Ammonia, toluene, formaldehyde, nitrogen dioxide	[219]

^a S = R_a/R_g. ^b R% = ((V_a − V_g)/V_g) × 100. ^c R_s = (V_work − V_output)/V_output × R_ref. ^d S = R_g/R_a. ^e S = (R_g − R_a)/R_a. ^f R_s = (I_g − I₀)/I₀. ^g S = (ΔI/I_a) × ^h R_s = (R_a − R_g)/R_a × 100. ⁱ S = (G_f − G₀)/G₀ = ΔG/G₀. ^jResponse = (G_air − G_gas)/G_gas. ^kRT—room temperature. ^lNA—not available.

Nanoplates of NiO/ZnO heterostructures were used to detect ethanol [65]. The morphology of the nanoplates of pristine NiO and NiO/ZnO is uniform, with a hexagonal shape and diameter of 90–120 nm, as shown in figures 8(a) and (b). The selectivity towards ethanol is indicated in figure 8(c) where data are presented for 500 ppm of oxidizing and reducing gases. Figure 8(d) shows the transient response-recovery to ethanol gas at an operating temperature of 400 °C, within the detection range from 10 to 800 ppm. The superior performance of NiO/ZnO heterostructure-based gas sensor is confirmed in figure 8(e), which is attributed to the higher catalytic activity compared to pure oxides and the presence of more active sites for the adsorption–desorption process [65]. Figure 8(f) illustrates the stability and reproducibility of the sensor made with NiO/ZnO.

Zoom In Zoom Out Reset image size

Morphology has been proven essential for sensing performance, as exemplified by comparing sensors fabricated with the same nanomaterials but distinct morphologies. NiO nanorods modified with ZnO nanoparticles were used to detect ethanol vapor at 100 ppm at 300 °C [37], while hierarchical hollow NiO/ZnO microspheres exhibited a higher sensing response at the optimum operating temperature of 260 °C [64]. Another SnO₂/ZnO p–n heterostructure was used for selective ethanol sensing, with a sensor response of 18.1 and fast response (3 s) at the operating temperature of 275 °C [66]. Therefore, in various studies the detection experiments are accompanied by a full characterization of the nanomaterials. Figures 9(a)–(d) show SEM and TEM images of p–n hollow nanospheres with n-type α-Fe₂O₃ nanoparticles on p-type LaFeO₃ [60]. The average diameter of the nanospheres was 350 nm, with the polycrystalline nature of the α-Fe₂O₃/LaFeO₃ nanospheres being confirmed with selected-area electron diffraction (SAED) in figures 9(e) and (f). Ethanol detection could be made in the range from 5 to 200 ppm at an operating temperature of 240 °C with excellent reproducibility, selectivity, and long–term stability (for 15 d), ultrafast response time (1 s), and recovery time (5 s) to 100 ppm ethanol (figures 9(g)–(k)).

Zoom In Zoom Out Reset image size

The detection of VOCs may also serve a disease diagnosis, for these compounds are present in humans exhaled breathing. For instance, acetone vapor from the human breath is associated with diabetes, which has motivated work on mesoporous hollow Zn₂SnO₄/SnO₂ microboxes whose sensing results are shown in figure 10 [69]. The porous microboxes with an interplanar spacing of 0.335 nm and 0.261 are shown in figures 10(a)–(e), which also include elemental mapping. The results of pore diameter, porosity, and surface area are given in figure 10(f), while the data related to sensing appear in figures 10(g)–(j). It is worth noting that the sensing performance was considerably higher (8–18 times) for the Zn₂SnO₄/SnO₂ than pure SnO₂ at the operating temperature of 250 °C.

Zoom In Zoom Out Reset image size

Other p–n heterostructures to detect acetone, ethanol, formaldehyde (HCHO), NO_2, and benzene (C₆H₆) include NiO/ZnO, SnO₂/NiO, Co₃O₄/ZnO and CuO/ZnO [82, 148, 211–215]. In all cases, the sensing performance is enhanced compared to pristine materials. There are, however, still limitations related to selectivity and operation temperature. Efforts have been made to mitigate such limitations, with metal doping and noble metal decoration to exploit catalytic properties and spillover effects. SnO₂/Au-doped In₂O₃ core–shell electrospun NFs were selective to acetone with excellent response/recovery speed (2 and 9 s) and stable performance [79]. As for VOCs, an essential factor for sensing performance is the bond dissociation energy [12, 220–225], which is temperature dependent and affects the optimum operating temperature. Morphology of the heterostructures is also important, as is the case for sensing the toxic gases NO₂, H₂S, and ammonia.

3.2. n–n heterostructures

The formation of n–n heterostructures induces band bending as in p–n heterostructures but the depletion layer appears owing to transfer of electrons from higher to lower energy levels. These heterostructures have been used for gas sensors, including TiO₂/SnO₂ [81, 226], CeO₂/SnO₂ [103] and SnO₂/ZnO [227–230], as listed in table 4. Sensing performance can be enhanced with oxygen vacancies serving as active sites on the surfaces and interfaces of these heterostructures. Moreover, they tend to be mesoporous which is beneficial to adsorption and desorption of gas molecules, as exemplified with n–n core–shell TiO₂/SnO₂ electrospun NFs to detect acetone gas [81]. The latter authors also investigated similar NFs for comparison purposes. Figures 11(a)–(c) shows a FESEM image of electrospun NFs from TiO₂–SnO₂ core–shells, SnO₂, and TiO₂ core–shells. No broken NFs are observed. The NFs are uniformly deposited as indicated in the TEM images of figures 11(d)–(f). Optimized sensing performance occurred at 280 °C with TiO₂–SnO₂ for 100 ppm acetone in figure 11(g). The selectivity towards acetone is depicted in figure 11(h), which may not be sufficient for practical applications. Figure 11(i) shows the transient response behavior from concentration 10–1500 ppm to acetone gas, with no saturation after exposure to 500 ppm. Sensors for TEA also had operating temperatures around 260 °C [226], while n–n heterostructures with CeO₂/SnO_2, SnO₂/ZnO, SnO₂/Zn₂SnO₄ have been used to detect H₂ and ethanol [66, 103, 231].

Zoom In Zoom Out Reset image size

Table 4. Gas sensing properties of n–n heterostructures.

Analyte gas	Morphology	Composition	Operating temperature	Concentration	Sensor response	Response/recovery time	LOD	Selective against	Reference
TEA	Nanoparticles—nanosheets	TiO2/SnO2	260 °C	100 ppm	52.3 a	12 s/22 s	2 ppm	Ethanol, glycol, isopropanol, paraxylene, acetone, benzene	[226]
Acetone	Core–shell NFs	TiO2/SnO2	280 °C	100 ppm	13.7 a	2 s/60 s	10 ppm	Acetone, formaldehyde, methanol, ethanol, toluene, ammonia	[81]
Hydrogen	Nanoparticles	CeO2/SnO2	300 °C	60 ppm	∼1323 a	17 s/24 s	0.5 ppm	CO, CH4, NH3, NO2, ethanol	[103]
Ethanol	Nanorods	SnO2/ZnO	275 °C	100 ppm	18.1 a	3 s/38 s	1 ppm	Xylene, toluene, formaldehyde, methanol, acetone	[66]
Ethanol	Porous spheres	SnO2/Zn2SnO4	250 °C	100 ppm	30.5 a	2 s/114 s	0.5 ppm	Acetone, methanol, formaldehyde, toluene, xylene	[231]
Hydrogen	Thin Films	SnO2–TiO2	400 °C	20 ppm	0.53 b	12–14 s/4.5–5 min	1 ppm	NA f	[228]
Trimethylamine	Hierarchical	Fe2O3/TiO2	250 °C	50 ppm	13.9 a	0.5 s/1.5 s	10 ppm	C7H8, HCHO, NH3, C3H6O	[180]
Ethanol	NFs	Nb2O5–TiO2	250 °C	500 ppm	21.64 a	NA f	0.017 ppm	Isopropyl alcohol (C3H8O), acetic acid (CH3COOH), ammonia (NH3), methylbenzene (C7H8), nitrogen dioxide (NO2)	[229]
Ethanol	NFs	TiO2/V2O5	350 °C	500 ppm	24.6 a	6 s/7 s	20 ppm	Acetone, ammonia, methanol, toluene	[232]
Ethanol	NFs	ZnO–TiO2	320 °C	50 ppm	10.6 a	5 s/10 s	20 ppm	H2S, CH4, C3H6O, CO, CH3OH, C2H2	[146]
Ethanol	Nanorods	MoO3/TiO2	180 °C	10 ppm	4.8 a	40 s/40 s	10 ppm	H2, NH3, CH4	[233]
Ethanol	Nanowires	SnO2/ZnO	400 °C	100 ppm	6.2 a	NA f	25 ppm	NH3, CO, H2, CO2, LPG	[227]
NO	NFs	ZnO/CdO	215 °C	33 ppm	22.6 a	35 s/630 s	1.2 ppm	NO2, H2S, CH4, SO2, CO, ethanol, acetone, ammonia, methanol, chloroform	[234]
Acetone	Nanoplates	ZnO/CdO	30 °C	100 ppm	540 c	61 s/47 s	1 ppm	Formaldehyde (HCHO), ethanol (C2H5OH), benzene (C6H6), ammonia (NH3)	[63]
NO2	Microspheres	SnO2–Sn3O4	150 °C	50 ppm	940 d	25 s/14 s	20 ppb	Ethanol, acetone, toluene, xylene, H2, CO, CH4	[61]
NO2	Hierarchical nanostructures	SnO2–ZnO	150 °C	10 ppm	106 d	50 s/55 s	5 ppb	Acetone, C7H8, H2S, CHCl3, NH3, ethanol	[50]
NO2	Heterojunction	ZnO/In2O3	RT	3 ppm	1419.9 e	13/189 s	500 ppb	H2S, NH3, SO2, acetone and ethanol	[207]
H2	Nanotubes	SnO2–decorated TiO2	250 °C	1000 ppm	1410	NA f	NA f	NH3, CH4, C3H8, and NO2.	[235]
Ethanol	Hierarchical structure	SnO2/ZnO	260 °C	100	366	8/45	NA f	Methanol, Acetone, Toluene, Benzene, Ammonia, TEA	[142]
NO2	Composite	ZnO/In2O3	RT (UV-LED)	5	2.21	100/695	NA f	NA f	[127]
Acetone	Hierarchical	Fe2O3/SnO2	250 °C	100	16.8	7/54 s	NA f	Ethanol, Methanol, HCHO	[153]

^a S = R_a/R_g. ^b S = (R_a − R_g)/R_g × 100. ^c S = (I_g − I_a)/I_a × 100. ^d S = R_g/R_a. ^e S = (R_g − R_a)/R_a × 100. ^fNA—not available.

The sensing mechanism for n–n heterostructures is illustrated in figure 12(a) for TiO₂/SnO₂ core–shells, where electrons from TiO₂ flow to the SnO₂ layer. This increases the number of electrons in the shell to be captured by adsorbed oxygens, thus enhancing the sensing performance in comparison to devices made with pristine SnO₂. Figure 12(b) shows a diagram for the energy band of the SnO₂/ZnO n–n heterojunction in air and in ethanol gas. Electrons flow from SnO₂ to ZnO to equalize the Fermi levels and create a depletion layer, leading to higher resistance of the composite material. Thus, when exposed to a reducing gas, the trapped electrons are released to the conduction band to increase sensor response.

Zoom In Zoom Out Reset image size

3.3. p–p heterostructures

Gas sensors have been increasingly made from 2D nanomaterials such as graphene and transition metals dichalcogenides (MoS₂, WS₂) due to their outstanding properties, namely RT operation and accelerated electron transfer [4, 34, 240–245]. These features are desirable since most chemiresistive sensors are operated at high temperatures, with risk of sensor degradation and explosion. A list of architectures and materials with these properties can be found in [201]. However, pristine 2D nanomaterials have poor gas sensing performance [90, 246, 247]. The low selectivity/sensitivity of rGO is explained by the inherent stacking tendency and periodic honeycomb lattice carbon structure, leading to adsorption of any molecule. The sluggish recovery time is assigned to remnant carboxylic acids that increase the binding energy [247]. Pristine MoS₂ strongly adsorbs atmospheric oxygen, which affects their activity and stability [246]. Various strategies are used to overcome these limitations, including external stimuli (heat, light) [242, 247–250], doping with noble metals [251, 252], exploring novel materials [253–258], and tailoring heterostructures by decorating with metal oxides. The last two strategies are the most attractive due to the myriad of possibilities to improve gas-sensing performance. Some examples exploiting heterojunctions with room- or low-temperature operation are discussed below.

The restacking problem was addressed by using single- and multiple-component crumpled 2D metal oxides, and graphene oxide as a sacrificial template [83]. As depicted in figure 13(a), the precursors were mixed with GO solution in a magnetic stirrer and the solution was nebulized through ultrasonic transducers in a tubular quartz reactor at 400 °C to promote shrinkage and compression of GO sheets. Crumpled heterostructures of GO/SnO₂, GO/ZnO, and GO/Co₃O₄ display uniform submicron size as shown in figures 13(b)–(d). The effects from the crumpled morphology and heterojunctions on gas sensing properties were verified with gold electrodes modified with 0D SnO₂, 2D SnO₂, crumpled 2D SnO₂, crumpled 2D SnO₂/ZnO, and crumpled 2D SnO₂/Co₃O₄. The data for the detection of acetone and formaldehyde are displayed in figures 13(e) and (f), where the highest sensor responses are observed for the crumbled heterostructures. A sensor array consisting of crumpled 2D SnO₂, crumpled 2D SnO₂/ZnO, and crumpled 2D SnO₂/Co₃O₄ was used to analyze selectivity. When the combined data are visualized using principal component (PCA) in figure 13(g), one notes that formaldehyde and acetone are discriminated completely without any overlap with other interferences or gas analytes. Although sensitivity and selectivity issues were mitigated, the operational temperature of crumpled 2D metal oxides was still high (350 °C–400 °C) [83].

Zoom In Zoom Out Reset image size

Nanotubes of TiO₂ were loaded with rGO to detect H₂ gas, where the increased conductance upon adding rGO led to an enhanced response compared to the sensor with pristine TiO₂ [259]. An even higher performance was obtained by using niobium-doped titanium dioxide (Nb–TiO₂) nanotubes [260]. rGO was also used with cerium oxide (CeO₂) nanocrystals to detect NO₂ at RT but recovery was only reached under light exposure at λ = 365 nm [247]. The sensitivity of rGO/CeO₂ heterostructures toward NO₂ was increased by exploring the effect of oxygen vacancies on CeO₂ nanoparticles via increasing the proportion of Ce³⁺ ions [261]. The enhancement was explained with first-principle calculations which indicated that CeO₂ becomes metallic at a given proportion of Ce³⁺ ions [261]. Detection of NO gas at 60 °C was reached with SnO₂ nanoparticles and Ni-decorated natural cellulosic graphene nanohybrid films (SnO₂/Ni-NCG) [262]. Although the sensitivity is not the highest in the literature, the operating temperature is the lowest.

ZnO is one of the most studied metal oxides in gas sensors, but their commercial application has been hampered especially owing to the requirement of high-temperature operation. This limitation has been addressed with heterostructures from oxide nanostructures and 2D materials. For example, a ternary Pt/rGO/ZnO heterostructure was produced to detect H₂ gas in the dark at 100 °C [42]. RT detection of H₂ was made possible by taking advantage of plasmonic properties from gold nanoparticles with rGO and ZnO nanorods [41]. Figures 14(a)–(f) show SEM images of the various nanostructures, while the sensing performance of ZnO, rGO/ZnO and Au/rGO/ZnO is depicted in figures 14(g)–(i), including under UV irradiation. As expected, the performance of the ZnO nanorod gas sensor is highly dependent on the temperature. The response is further improved to 35% with the rGO/ZnO nanocomposite, which decreased the optimal temperature to 150 °C under UV irradiation. The most significant improvement was observed for the Au/rGO/ZnO nanocomposite that reached 96% with fast response/recovery times of 8/612 s at RT under UV light. It is worth mentioning that a long-term experiment of Au/rGO/ZnO kept the response approximately constant for over 60 d. The proposed H₂ sensing mechanism is schematically as follows: when the sensor is exposed to ambient air, oxygen molecules are adsorbed on the sensor surface, making the electrical resistance increase due to the decrease in the carrier charge density depletion layer (figure 14(j)). When exposed to UV light under an air atmosphere as in figure 14(k), abundant electrons and holes are generated on the sensor surface, increasing the depletion region width and decreasing the electrical resistance. The exposure of the Au/rGO/ZnO sensor to H₂ gas represented in figure 14(l) leads to water as a product of ionized oxygen species and H₂ molecules reaction. This decreased the electrical resistance due to the release of trapped electrons back to the bulk of the sensor [41].

Zoom In Zoom Out Reset image size

The discrimination of similar molecules remains a challenge for metal oxide gas sensors [263]. This can be achieved using the concept of electronic noses according to which the global selectivity principle with a combination of various sensing units is applied to solve problems related to selectivity. A multisensory system comprising three heterostructured materials was prepared via a one-step hydrothermal route and layer-by-layer (LbL) technique, as shown in figure 15(a) [264]. Tin dioxide (SnO₂) nanospheres and copper oxide (CuO) nanoflowers-decorated graphene were used as active layers for sensors and deposited on Cu/Ni interdigitated electrodes. The I–V measurements collected at RT as depicted in figure 15(b) led to different profiles depending on the gas and on the sensing unit, which were analyzed with a machine learning algorithm. Figure 15(c) illustrates the data processing procedure with which a quantitative prediction of formaldehyde (HCHO) and ammonia (NH₃) in gas mixtures was obtained using a backpropagation neural network model. The adequacy of the machine learning model was confirmed by noting the high correlation between the predicted and measured concentrations in figure 15(d). This kind of machine learning model is widely used for pattern recognition and quantitative prediction in many fields [265–272].

Zoom In Zoom Out Reset image size

MoS₂ nanosheets are being considered as an alternative material to graphene that has zero bandgap structure, with MoS₂ bandgap ranging from 1.5 to 3 eV [273]. The activity and stability of pristine MoS₂ are strongly hindered by atmospheric oxygen adsorption, though. Elegant solutions to this problem consist of metal doping, using heterojunctions, and decoration with metal oxides. Metal doping provides high sensitivity [274, 275], while high-entropy alloys (HEA) such as Ag, Au, Pt, Pd, Cu are relatively cheaper. MoS₂ sheets decorated with HEA nanoparticles were used to detect H₂ gas at 80 °C, with a similar performance to MoS₂ sensors doped with pure metal nanoparticles [276]. Cu₂O–MoS₂ microspheres were utilized for ammonia sensing at 130 °C [277], and RT detection was achieved with MoS₂/NiO heterostructures [90]. Also possible to operate at RT was the NO₂ gas sensor made with MoS₂@MoO₂ n–p heterojunctions [246], and the ethanol vapor sensor fabricated with multilayered hierarchical α-Fe₂O₃/MoS₂ heterostructures [278]. The main significance of the latter work was in the ultrafast response of an ethanol sensor operating at RT, which did not require sophisticated preparation methods. It is worth mentioning that sensing performance for the same materials may depend on the fabrication procedures and conditions. For instance, the thickness of the MoS₂ layer in MoS₂/ZnO heterostructures was an important factor detecting NO₂ [279]. Whether a RT operation is made possible also depends on the fabrication method. This was reached for NO₂ detection with MoS₂/ZnO heterostructures prepared either with a chemical precipitation process [249] or a wet chemical method [67].

The 2D materials referred to as MXenes are transition metal carbides, nitrides, and carbonitrides with the general formula M_{n +1} X_nT_x , where M is an early transition element (Ti, Nb, V, Cr, Mo, Ta), X are layers of C and/or N, T_x represents the surface termination groups (–OH, –O, –F) and n varies from 1 to 4 [280]. They have been used in various gas sensors, as in the case of pristine Ti₃C₂T_x MXene whose signal-to-noise ratio was considerably higher than other 2D materials [281, 282]. However, aggregation is still an issue comprising their sensitivity and selectivity. A flexible sensor for NO₂ was produced from pristine 3D crumpled MXene Ti₃C₂T_x spheres and a Ti₃C₂T_x /ZnO composite via ultrasonic SP, as depicted in figure 16(a). The crumpled nature of MXene and composite structures is confirmed in the SEM images in figures 16(b) and (f), while maps of elemental analysis are displayed in figures 16(c)–(e) and 16(g)–(j) [283]. It is important to remark that Ti₃C₂T_x MXenes are adequate for sensing NO_x species, but not for some VOCs and ammonia, which can be addressed by combining MXenes with CuO [284].

Zoom In Zoom Out Reset image size

The articles mentioned above are demonstration of the advances in gas sensing through novel heterostructures that operate at a room- or low-temperature, thus addressing some of the challenges hampering commercialization. There is nevertheless plenty of room for improvements to reach gas sensors that operate at RT and detect multiple analytes without suffering from interfering issues. To fulfill these stringent requirements, we suggest that efforts should be made to develop gas sensor arrays combined with artificial intelligence methods for data analysis.

We have already mentioned that gas sensors made with metal oxides are usually operated at high temperatures. In order to reach detection at the desired RT, researchers have used various strategies, including surface modification using noble metal nanoparticles or metal/metal oxide doping [70, 71, 209]. Table 5 summarizes the details of RT sensors to detect toxic gases with MOHs. For example, Al-doped ZnO/CuO nanocomposites were used to detect ammonia selectively and at RT, with fast response (14 s) and recovery (9 s) [68]. NO₂ sensors produced from SnO/SnO₂ p–n heterostructures could operate at RT and high sensitivity, but the recovery speed was poor (5 min) [120]. Another important factor for RT sensors is humidity, for sensors have to maintain their performance under varying humidity conditions. This has been achieved with H₂S sensing at 70 °C and relative humidity from 11% to 95% using CuO/In₂O₃ heterostructures with MOFs [73], and with ammonia detection at RT using tin–titanium dioxide/reduced graphene/carbon nanotube (Sn–TiO₂@rGO/CNT) composites [74]. Sensing performance at RT can also be enhanced via illumination or doping with 2D materials. Indeed, MoS₂/ZnO hetero-nanostructures were employed to detect NO₂ at low ppb concentrations at RT [67].

In this review paper, we have so far concentrated only on the challenges and opportunities to exploit MOHs in gas sensors, with no emphasis on their possible integration into other devices and monitoring systems. We now highlight the prospects of using these materials in sensors that may exhibit special characteristics, such as being self-powered. This topic has grown rapidly with piezoelectric and triboelectric nanogenerators, including gas sensing properties [285–290]. For example, a selective self-powered RT H₂S sensor was obtained with the NiO/ZnO nanowire arrays in figure 17(a) [209]. The sensor device was connected with a low-noise preamplifier to monitor the piezo-voltage response upon exposure to H₂S gas. Figures 17(b) and (c) show the selectivity histogram of the sensor for toxic gases individually and in a mixture. The piezo-voltage response in figure 17(d) was ten times the value for pure ZnO, with a low detection limit of 10 ppm. Other examples of heterostructures utilized in self-powered room RT gas sensors are described in [1, 289, 290].

Zoom In Zoom Out Reset image size

Sensing performance is affected by various environmental factors and device characteristics, such as humidity, temperature, film thickness, electrode area, and particle size/shape [315]. The grain size and width of the necks, in particular, play a major role [316, 317], which can be understood by considering the model depicted in figure 18, where different parameters are proposed to control the properties. Smaller grain sizes offer high surface area and, therefore, higher sensitivity, while larger grain sizes lead to less homogeneous and stable materials. For larger grains where D (grain size) ≫ 2 l, grain size is unaffected by the surface interaction, and resistivity is dominated by the grain boundaries. When D = 2 l, grain size decreases in the depletion region, and the necks become more resistant (neck control). If D < 2 l, the depletion region extends to the whole area of grains and the resistivity is governed by the grain size effect (grain control) [318]. The Debye length (L_D) can be related to the grain size (d) or neck width as in equation (2):

$$\begin{equation}{L_{\text{D}}}{\mkern 1mu} = {\mkern 1mu} \sqrt {\frac{{{\varepsilon _{\text{s}}}{\mkern 1mu} kT}}{{{q^2}{N_{\text{d}}}}}} {\mkern 1mu} \end{equation} \tag{ 2 }$$

Zoom In Zoom Out Reset image size

where T is the absolute temperature, ε is the dielectric constant of the material, k is the Boltzmann constant, and N is the number of charge carriers. The grain size affects the gas sensing properties, as observed with acetone sensing at ppb level with core–shell ZnO–CuO p–n heterojunctions, where performance increased with small grain sizes and large surface areas [319]. Another approach to enhance sensitivity is to control morphology and porosity of oxide heterostructures [320].

Another critical factor for sensing performance is humidity, since an increased humidity may decrease the sensor response because chemisorbed water molecules will diminish the adsorbed oxygen species following the Grotthuss mechanism [321, 322]. Because heat treatment at 400 °C could desorb these water molecules, most metal oxides work at high temperatures to achieve high sensor response [323, 324]. This also illustrates the importance of the operating temperature [20, 325–327]. At low temperatures, the reaction rate is slow and the sensor response is relatively small. Upon increasing the temperature, the response gradually increases because of exothermic reactions between gas molecules and adsorbed oxygen species. However, when the temperature increases above a threshold value, the desorption of chemisorbed oxygen is enhanced to a higher rate than the reaction rate. The activation energy of the gas is not sufficient to cross over the conduction band, thus leading to a decrease in response. Strategies to overcome the problems related to the operation temperature include the use of a heating platform to save power consumption, in addition to surface modification and metal doping [328].

Film thickness and effective area of the sensing units may also affect the performance. For instance, reducing film thickness from 600 to 50 nm led to a five-fold increase in the response of SnO₂ sensors to ethanol, which was explained by the easier penetration of gas molecules to enhance the adsorption–desorption process [329, 330]. It is also relevant that reducing film thickness makes it possible to reduce power consumption and increase the measurement accuracy [329, 331].

In this review, we have summarized the recent advances in gas sensors made with MOHs, including gas sensing mechanisms and the search for enhancing sensing performance for a variety of gases. Interest in these heterostructures arises from their large specific surface area, low cost, facile fabrication and fast adsorption–desorption towards target gases. Applications of such gas sensors encompass monitoring systems in the health and environment sectors, in many cases with the performance using heterostructures being higher than that obtained with the corresponding pristine materials. Architectures based on n–n, n–p, and p–p semiconducting oxides have been used in chemiresistive gas sensors. Also discussed in the review were the fabrication methods which include solvothermal, hydrothermal and electrospinning techniques, with which different shapes and morphologies could be obtained for tailoring the sensing properties. The main challenge with regard to fabrication methods is the need of upscaling for the various heterostructures such as core–shell, layered, and 1D–1D, 1D–2D, and 2D–2D heterostructures. Fortunately, these heterostructures typically have a lower optimum working temperature than their pristine materials, which allows to partially mitigate the diffusion mechanism [25].

Upon surveying the limitations of the heterostructures for gas sensors, we identified those mainly related to the operation temperature, selectivity and stability. Strategies to improve selectivity of these heterostructures include: (i) incorporating suitable additives, (ii) temperature control, (iii) using appropriate filters, (iv) UV-light illumination [332]. Materials should be selected which possess high catalytic activity against the target gas, and complex structures should be designed without affecting the properties of the heterostructures, such as surface area, porosity, electrical conductivity. Noble metals such as gold and palladium, for example, have catalytic activity against CO and H₂ gas [333, 334], which can be increased upon controlling morphological parameters such as size and shape. The combination of metal oxides and 2D materials such as MXenes or conducting polymers is promising, but the sensing mechanism has not been determined accurately. Sensing efficiency can be achieved by electronic and chemical sensitization to regulate charge carrier concentration with electrons being transferred from metal ions to MOHs. Such heterostructures should therefore be designed with materials with different work functions. The selectivity problem may be overcome by using a set of microsensor arrays, particularly with the use of advanced data analysis methods, including machine learning.

In spite of the difficulties mentioned about the operation temperature, many examples exist of RT sensors. Operation under high humidity is still a serious issue, as is the response time. The power consumption problem may be solved by miniaturizing the sensors using microfabrication (microelectromechanical systems) techniques. These micro-level sensors can be fabricated by incorporating hotplates with a sensitive sensing element. Since humidity affects the operating temperature, it is essential to develop a gas sensing material that resists humidity interference. MOFs can provide new perspectives for interference for these sensors.

Another avenue to pursue is to fabricate flexible, wearable devices, especially for healthcare [335–338]. Among the challenges in this topic is the use of polymer or cellulose materials in the substrates [1], which is restrictive for most metal oxide materials because of their low glass transition temperature. This requires another synthesis strategy with a low-temperature chemistry method. For example, one may post-transfer the nanomaterial to the flexible electrodes by drop-casting [283], preferably at RT. In addition, the sensing material must endure mechanical strain without affecting the sensing performance [339]. As for scaling up devices, screen-printing with nanocomposite inks may be a viable alternative. The search for the solution to these problems offers a myriad of possibilities.

As a conclusion, we take the view that MOHs may have a tremendous impact on chemical sensing. This will require experimental and theoretical work to determine sensing mechanisms and exploit new materials such as transition metal dichalcogenides, MXenes, black phosphorous, and other carbon materials.

This work was carried out with financial assistance from the Brazilian funding agencies: São Paulo Research Foundation-FAPESP (2014/23546-1, 2015/14836-9, 2016/23474-6, 2018/22214-6) and National council for scientific and technological development—CNPq (308570/2018-9).

No new data were created or analyzed in this study.

All the authors have contributed equally to this work.

The authors declare no conflict of interest.